PD-L1 expression in anogenital and oropharyngeal squamous cell carcinomas associated with different clinicopathological features, HPV status and prognosis: a meta-analysis

Abstract Background: Little research has been done on clinicopathological characteristics and human papillomavirus (HPV) status of anogenital and oropharyngeal squamous cell carcinomas (SCC) with a strong expression of programmed death ligand 1 (PD-L1) in tumor cells. Therefore, we conducted this meta-analysis. Methods: We performed a comprehensive research in PubMed, Embase and Cochrane databases up to 30 September 2020. The effect size was hazard ratio (HR) with 95% confidence interval (CI) for overall survival (OS), cancer-specific survival (CSS), disease-free survival (DFS). The pooled odds ratio (OR) with 95% CI were used to assess the association between PD-L1 expression and clinicopathological features along with HPV status. Results: A total of 2003 cases (944 anogenital and 1059 oropharynx SCC patients) were included. High PD-L1 expression in anogenital SCC cases were associated with advanced age (OR = 1.63, 95% CI: 1.04–2.58) and HPV negativity (OR = 0.47, 95% CI: 0.31–0.71). Besides, PD-L1 positive anogenital SCC cases held a significantly declined OS (HR = 2.18, 95% CI: 1.37–3.47) and CSS (HR = 2.45, 95% CI: 1.30–4.65). For oropharynx SCC, PD-L1 was more frequent in younger and HPV positive patients (OR = 0.60, 95% CI: 0.37–0.98; OR = 3.01, 95% CI: 1.78–5.09) and PD-L1 expression was relevant to better OS and DFS (HR = 0.76, 95% CI: 0.60–0.97; HR = 0.50, 95% CI: 0.33–0.75). Conclusions: The meta-analysis demonstrated that in anogenital SCC, PD-L1 positivity had to do with a worse outcome, which might attribute to advanced age, higher tumor grade, lymph node metastasis and HPV negativity, while in oropharynx cancer, PD-L1 expression was related to better prognosis for the reason that PD-L1 was less frequent in the aged and negative HPV status.


Introduction
Human papillomavirus (HPV)-related cancers account for 8.6% of female and 0.8% of male carcinomas globally, and persistent high-risk HPVs infection is the fundamental reason. High-risk HPVs not only contribute to more than 80% of cervical cancer cases, but also take responsibility for approximately 88% of anal, 78% of vaginal, 51% of penile, 40% of vulvar and 30% of oropharynx cancer cases. Besides, these malignancies account for almost all of HPV-related cancers [1]. Thanks to promoting screening programs and advances in treatment of cervical lesion, the incidence and mortality rates of cervical cancer generally have declined over the last decade. However, this tendency varies in different countries and regions [2]. Among men, approx. 3200 new anal cancer along with 26000 new penile cancer cases occur annually.
Meanwhile, approximately 4850 new anal cancer in 2016, 6190 new vulvar cancer in 2018 and 12000 vaginal cancer cases in 2012 were diagnosed among women worldwide [1,3,4].
In addition to anogenital cancers, head and neck cancer is the sixth leading cancer worldwide with 600000 new cases per year and a mortality rate of approx. 450000 per year, occupying 78% of HPV-related oropharynx cancer cases [1,5]. Meanwhile, the HPV positive oropharyngeal cancers have been increasing during the last decade, even though the consumption of tobacco declining, especially among youngsters in more developed countries [1]. Overall, early-stage HPV-related cancer patients have an optimal outcome with a high probability of cure. However, patients with advanced cancers, particularly with regional lymph nodes and distant metastases, may undergo multimodal treatments, including lymphadenectomy and chemoradiotherapy [6]. Unfortunately, these aggressive strategies can not prevent patients with high-stage carcinomas from treatment failure and poor prognosis. Therefore, new effective and optimized therapeutic options for those with advanced tumors are urgent.
Recently, the early-and late-phase clinical researches show that the efficiency of immunotherapies in certain solid tumors brings hope for patients, such as anti-programmed death 1 (PD-1)/programmed death ligand 1 (PD-L1) immunotherapy, which is one of the significant immunotherapies. And, PD-1 or PD-L1 blockade therapy is more effective than conventional therapies in advance solid tumors [7]. PD-1, a co-inhibitory receptor existing on T cells, B cells and natural killer (NK) cells, is combined with its ligand PD-L1 in tumor cells, which suppresses the activation and proliferation of T cells. As a result, tumor cells can escape from host anti-tumor immunity. By stopping the PD-1/PD-L1 interacting with PD-1 or PD-L1 blockade therapy, cytotoxic T cells are re-stimulated to eliminate cancer cells [8]. There are many kinds of cancers expressing PD-L1, including HPV-related malignancies.
The studies investigating the relationship between PD-L1 expression in tumor cells and clinical features in anogenital and oropharyngeal squamous cell carcinomas (SCCs) are growing , which could identify potential pathological characteristics for judging whether patients could get clinical benefits from PD-1 or PD-L1 blockade therapy with variable PD-L1 expression and investigate the role of PD-L1 in anogenital and oropharyngeal SCC. However, the associations between PD-L1 expression in anogenital and oropharyngeal SCC and clinicopathological features, HPV status (i.e. based on HPV DNA, p16 overexpression or both [30]) and prognosis are inconsistent. Therefore, we conducted the meta-analysis to clarify the association between PD-L1 expression in cancer cells and outcome, and to assess the relationship between PD-L1 expression in tumor cells and clinicopathological features along with HPV status in anogenital and oropharyngeal SCC.

Literature search
A comprehensive research was conducted in PubMed, Embase and Cochrane databases based on 'PRISMA' guidelines up to 30 September 2020. Furthermore, there was no language restriction. The following items were used for searching: ('programmed cell ligand 1' or 'PD-L1' or 'B7-H1' or 'CD274') and ('HPV-related' or 'penile' or 'penis' or 'cervical' or 'cervix' or 'vaginal' or 'vagina' or 'vulvar' or 'vulva' or 'anal' or 'anus' or 'oropharyngeal' or 'base of tongue' or 'posterior pharyngeal wall' or 'tonsil' or 'soft palate') and ('cancer' or 'tumor' or 'neoplasm'). Furthermore, the references of included articles were also manually screened for other potential eligible researches.

Inclusion and exclusion criteria
Eligible studies had to fulfill the following criteria: (1) all SCC were diagnosed by pathology results.

Data extraction
The relevant data were extracted from eligible studies by two independent authors (X.Z.) and (J.c.L.), which were examined by other investigators (Y.L. and Y.Q.). The data extracted from the eligible studies were recorded as follows: first author, year of publication, cancer type, country of patients population, gender, age, follow-up period, patients sample size, pathology, methods for PD-L1 detection, PD-L1 distribution, cut-off value of PD-L1 positive expression, PD-L1 positive expression rate, tumor stage and outcomes of patients, including overall survival (OS), cancer-specific survival (CSS), disease-free survival (DFS) with hazard ratios (HRs) with 95% confidence interval (CI). Moreover, Newcastle-Ottawa Scale was applied to assess the quality of included papers.

Statistical analysis
Stata software (version 12.0; Stata Corp LP, College Station, TX) was applied for present meta-analysis. The relationships between PD-L1 and clinicopathological characteristics along with HPV status were evaluated by odds ratio (OR) with 95% CI. If studies provided HPV-DNA and P16 status in papers, we considered that P16 was a superior biomarker to HPV-DNA due to P16 best representing oncogenic activity of the HPV in tumors [31]. HR with 95% CI directly reported in eligible studies was applied to assess the role of PD-L1 in prognosis of patients, and we also adopted method described by Altman and Bland to calculate 95% CI, when its HR and P-value were provided in study [32]. Moreover, if studies offered Kaplan-Meier curves rather than HR with 95% CI, we extrapolated HR with 95% CI by data directly obtained from the curves. When multivariable and univariable outcome analyses were offered, the former was adopted. Because of the obvious differences between anogenital cancers (cervical, vaginal, vulvar, penile and anal cancer) and oropharyngeal cancers in embryological origins, as well as unequivocal premalignant and malignant changes existing in entire anogenital region, we divided anogenital and oropharyngeal SCC in this meta-analysis into two main types, anogenital and oropharyngeal SCC, of which we performed meta-analysis, respectively.
Heterogeneity between researches was evaluated by Chi-square test and I-square test, and when value of P<0.10 or I 2 > 50%, we considered heterogeneity present in studies and utilized the random-effects model rather than fixed-effects model to calculate the pooled ORs/HRs. Otherwise, the fixed-effects model was applied with P>0.10 and I 2 < 50%. Additionally, we performed one-way sensitivity analysis to assess the stability of the results. Moreover, Egger's and Begg's tests were applied to evaluate the publication bias among the studies. Two-tailed P-values less than 0.05 were considered statistically significant.

Literature search
Initially, a sum of 570 articles were identified according to previous search strategy. Then, 328 duplications were excluded. After reviewing the title and abstract, 199 studies were removed for reasons. The full texts of the remaining articles were estimated, and 23 studies were further excluded. Eventually, we included 20 articles from the rest studies in total. A flow diagram detailing the selection process of studies was shown in Figure 1.

Sensitivity and publication analysis
We confirmed that there was low heterogeneity in studies in the present meta-analysis by sensitivity analysis. Moreover, no substantial asymmetry was identified by Begg's tests and funnel plot, in the light of the visual inspection of the shape (Supplementary Figures S1 and S2). This indicated low publication bias and outcomes of meta-analysis to be statistically robust.

Discussion
The PD-L1/PD-1 pathway has showed a remarkable value among all immune checkpoints because promising and impressing responses in many tumors, such as melanoma, non-small-cell lung cancer (NSCLC) and renal cell carcinoma, were achieved in therapies targeting the PD-1/PD-L1 pathway [33]. In HPV-related carcinomas, antibodies targeting the PD-1/PD-L1 pathway are being assessed in oropharynx cancers, cervical cancer, penile cancer and anal cancer [34][35][36]. A phase Ib clinical trial demonstrated that the overall response rate (ORR) of squamous cell carcinoma of the head and neck SCC cases receiving pembrolizumab was 18%, and complete response was 5% [37]. The anti-tumor activity of pembrolizumab was also shown in advance cervical cancer, whose ORR was 17% and treatment-related adverse events were acceptable [34]. Recently, a phase II trial of nivolumab in refractory metastatic anal SCC illustrated that the ORR was 21% [35]. Moreover, the clinical studies exploring immunotherapy for the treatment of penile cancer are ongoing [36]. Nevertheless, not all of the cases treated with anti-PD-1/PD-L1 therapy got clinical response. Therefore, it is necessary to identify biomarkers with therapeutic effect to select patients and predict response. PD-L1 expression in tumor cells was considered as a potential response biomarker for PD-1/PD-L1 targeted therapy in most studies, including the clinical trials of nivolumab/pembrolizumab in HNSCC, cervical and anal cancer, despite PD-L1 negative patients could also get clinical benefits from the therapy [7,34,35,37]. Hence, it is necessary to clarify the role of PD-L1 in anogenital and oropharyngeal SCC, which could help to identify more suitable anogenital and oropharyngeal cancer cases for anti-PD-1/PD-L1 therapy.
In the meta-analysis of anogenital SCC, high PD-L1 expression was related to advanced age and higher tumor grade. Furthermore, PD-L1 positive patients held a worse OS in all anogenital SCC, and a worse CSS in penile SCC. PD-L1, as the dominant inhibitory ligand of PD-1, could induce a conformational change of PD1 and weaken T cell-activating signals. Consequently, proliferation, survival, cytokine production and other functions of T cell were inhibited. Additionally, signaling roles of PD-L1 molecule were identified in certain studies. PD-L1 can deliver pro-survival signals to cancer cells, leading to resistance of apoptosis. Furthermore, PD-L1 can prevent tumor cells from immune cytotoxic effects without assistant of PD-1 signal in T cells [33]. Therefore, the anogenital SCC become more aggressive with high PD-L1 expression, such as higher tumor grade, lymph node metastases and worse prognosis.
In HNSCC cells, Interferon-γ (IFN-γ), produced by CD8 + tumor-infiltrated lymphocytes (TILs), was demonstrated to control the gene expression of PD-L1 in tumor cells [28]. Higher number of CD8 + TILs or TILs were discovered in HPV-positive HNSCC patients compared with HPV-negative cases [28,38]. As a result, accumulation of activated TILs and IFN-γ could explain favorable outcome in PD-L1 positive HNSCC patients, which were HPV positive.
However, IFN-γ, produced by TILs, seems not to be the main reason for the difference of PD-L1 expression between HPV positive and negative anogenital cancer patients, because no significant difference was identified in numbers of CD8 + TILs distributing in HPV positive and negative anal and cervical cancer patients [38,39]. In vulvar SCC, there was fewer number of CD8+ TILs in p16 positive cases compared with p16 negative [40]. Therefore, genetic background, including genomic aberrations and aberrant oncogenic signaling may take the primary responsibility for PD-L1 overexpression in anogenital cancers [17,41]. Thus PD-L1 was strongly associated with advanced age, lymph node metastases and worse outcomes in SCC.
In our meta-analysis, there was a distinct relationship between HPV status and PD-L1 expression in oropharyngeal and non-oropharyngeal tumors, especially penile SCC. And it also might explain the relationship between PD-L1 expression and clinical outcome, as HPV-associated oropharyngeal SCC had a better clinical outcome and non-HPV-associated anogenital SCC had a worse clinical outcome. Solomon et al. [38] found that oropharynx SCC patients with high PD-L1 expression seemed to hold a worse OS compared with low PD-L1 expression cases in P16 positive patients, although results were not statistically significant in multivariable analysis (HR = 1.9, 95% CI: 0.7-5.6). In addition to HPV status, the divergent relations between age of patients and PD-L1 expression could also explain that oropharyngeal SCC cases with high PD-L1 expression held a better prognosis, because PD-L1 in oropharynx SCC was more frequent in younger patients.
In order to reduce heterogeneous variables among the anogenital SCC studies, subgroups of meta-analysis were performed based on cancer types, antibody catalogs and cut-off values of PD-L1 positivity. Antibody (clone E1L3N and clone 22C3) and antibody (clone SP142) were reliable to detect PD-L1 expression in anogenital and oropharyngeal SCC, respectively. Moreover, it was rational that cut-off value of PD-L1 positivity was greater than or equal to 5%. Because of limited data of each cancer type in anogenital SCC, we could not perform meta-analysis of each kind of anogenital SCC. However, anogenital SCC (cervical, vaginal, vulvar, penile and anal cancer) had the same embryological origins, as well as unequivocal premalignant and malignant changes existing in entire anogenital region, so we considered that lumping all non-oropharyngeal tumors for meta-analysis was relatively rational.
Undoubtedly, there were some limitations in the meta-analysis. First, relatively fewer vulvar, penile and anal cancer patients were enrolled. Furthermore, no studies investigating the role of PD-L1 in vaginal cancer were found after research. Therefore, clinical studies with higher quality and large sample size are necessary to support our conclusion. Second, chemoradiotherapy also affects expression of PD-L1, but limited studies reported the detailed treatments of patients before examining the PD-L1 expression. As a result, we did not investigate the relationship between various forms of treatments and PD-L1 expression. Third, we estimated HR and 95% CI from Kaplan-Meier curves and with the method described by Altman and Bland [32], which were not accurate as reported by authors. Fourth, we could not judge whether different PD-L1 antibodies could lead to differing results for PD-L1 expression in HPV-related SCC due to the limited data involving antibody catalogs of PD-L1 in HPV-related SCC. Moreover, we should highlight the lack of analytical harmonization to PD-L1 evaluation, such as the consensus for p16 expression associated with HPV.

Conclusions
The results of the meta-analysis demonstrated that advanced age, higher tumor grade, lymph node metastasis and HPV negativity were associated with high PD-L1 expression in anogenital SCC cases. And anogenital SCC cases that were PD-L1 positive held a worse outcome. For oropharynx cancers, PD-L1 was less frequent in older patients and negative HPV status. Furthermore, PD-L1 expression was related to better prognosis of oropharynx cancer patients. Advanced age and negative HPV status might explain anogenital SCC cases with high PD-L1 expression holding a worse prognosis, compared with oropharynx SCC cases with high PD-L1 expression. Meanwhile, PD-L1 expression should be combined with clinicopathologic features representing high mutation load, including advanced age, higher tumor stage and HPV status, to identify more suitable HPV-related cancer cases for anti-PD-1/PD-L1 therapy.

Data Availability
All the data used to support the findings of the present study are included within the article. Please contact author for data requests.